PIMMS43 is required for malaria parasite immuneevasion and sporogonic development in themosquito vectorChiamaka V. Ukegbua,1, Maria Giorgallia,1, Sofia Tapanellia, Luisa D. P. Ronaa,2,3, Amie Jayea, Claudia Wyera,Fiona Angrisanoa,4, Andrew M. Blagborougha,4, George K. Christophidesa,5, and Dina Vlachoua,5

aDepartment of Life Sciences, Imperial College London, SW7 2AZ London, United Kingdom

Edited by Anthony A. James, University of California, Irvine, CA, and approved February 21, 2020 (received for review November 18, 2019)

After being ingested by a female Anopheles mosquito during abloodmeal on an infected host, and before they can reach themosquito salivary glands to be transmitted to a new host, Plasmo-dium parasites must establish an infection of the mosquito midgutin the form of oocysts. To achieve this, they must first survive aseries of robust innate immune responses that take place prior to,during, and immediately after ookinete traversal of the midgutepithelium. Understanding how parasites may evade these re-sponses could highlight new ways to block malaria transmission.We show that an ookinete and sporozoite surface protein desig-nated as PIMMS43 (Plasmodium Infection of the Mosquito MidgutScreen 43) is required for parasite evasion of the Anopheles coluzziicomplement-like response. Disruption of PIMMS43 in the rodentmalaria parasite Plasmodium berghei triggers robust complementactivation and ookinete elimination upon mosquito midgut traversal.Silencing components of the complement-like system through RNAilargely restores ookinete-to-oocyst transition but oocysts remainsmall in size and produce a very small number of sporozoites thatadditionally are not infectious, indicating that PIMMS43 is also essen-tial for sporogonic development in the oocyst. Antibodies that bindPIMMS43 interfere with parasite immune evasion when ingestedwith the infectious blood meal and significantly reduce the preva-lence and intensity of infection. PIMMS43 genetic structure acrossAfrican Plasmodium falciparum populations indicates allelic adapta-tion to sympatric vector populations. These data add to our under-standing of mosquito–parasite interactions and identify PIMMS43 as atarget of malaria transmission blocking.

malaria transmission | mosquito innate immunity | complement-likeresponse | transmission blocking vaccines | mosquito populationreplacement

Enhanced vector control significantly reduced malaria cases inrecent years and, together with effective medicines and better

health care, decreased the number of malaria-associated deaths.However, the effectiveness of these measures is currently com-promised due to widespread mosquito resistance to insecticidesused in bed-net impregnation and indoor residual spraying, whilemosquito biting and resting behaviors have also changed in re-sponse to these measures. As a result, no significant progress inreducing the global malaria burden is recorded in the past years.Therefore, additional tools for malaria control are needed, thedevelopment of which could be guided by a better understandingof disease transmission through the vector.Mosquito acquisition of Plasmodium parasites commences when

a female Anopheles mosquito ingests gametocyte-containing bloodfrom an infected person. In the mosquito midgut lumen, game-tocytes mature and produce gametes. Fertilization of gametesleads to zygotes that soon develop to ookinetes and traverse themidgut epithelium. At the midgut basal subepithelial space,ookinetes differentiate into replicative oocysts wherein hundredsof sporozoites develop within a period of 1 to 2 wk. Upon releaseinto the hemocoel, sporozoites—transported by the hemolymph—

traverse the salivary glands and infect a new host upon a nextmosquito bite.Inside the mosquito, parasites are attacked by an array of immune

responses (1, 2). Most parasite losses occur during the ookinete-to-oocyst transition (3, 4). Ookinete traversal of the mosquito midgutleads to activation of JNK (c-Jun N-terminal kinase) signaling, in-ducing apoptosis of the invaded cells. This response involves variouseffectors, including heme peroxidase 2 and NADPH oxidase 5 thatpotentiate nitration of ookinetes that are henceforth marked forelimination by reactions of the mosquito complement-like system(5, 6). These reactions are triggered upon ookinete exit at the

Significance

Malaria is transmitted among humans through mosquito bites.Here, we characterize a protein found on the surface of mos-quito stages of malaria parasites and reveal that it serves toevade the mosquito immune system and ensure diseasetransmission. Neutralization of PIMMS43 (Plasmodium Infectionof the Mosquito Midgut Screen 43), either by eliminating it fromthe parasite genome or by preincubating parasites with anti-bodies that bind to the PIMMS43 protein, inhibits mosquito in-fection with malaria parasites. Differences in PIMMS43 detectedbetween African malaria parasite populations suggest that thesepopulations have adapted for transmission by different mosquitovectors that are also differentially distributed across the continent.We conclude that targeting PIMMS43 can block malaria parasitesinside mosquitoes before they can infect humans.

Author contributions: C.V.U., G.K.C., and D.V. designed research; C.V.U., M.G., S.T., L.D.P.R.,C.W., F.A., and A.M.B. performed research; C.V.U., M.G., L.D.P.R., A.J., G.K.C., and D.V.analyzed data; C.V.U., M.G., L.P.D.R., G.K.C., and D.V. conducted methodology; C.V.U.,M.G., L.P.D.R., G.K.C., and D.V. conducted formal analysis; C.V.U. and M.G. contributed towriting the original draft; M.G. and C.W. conducted validation; A.J. and D.V. conducted datacuration; G.K.C. and D.V. conducted conceptualization; G.K.C. and D.V. conducted visualiza-tion, supervision, and project administration; and G.K.C. and D.V. wrote the paper.

The authors declare no competing interest.

This article is a PNAS Direct Submission.

This open access article is distributed under Creative Commons Attribution License 4.0(CC BY).

Data deposition: The RNA sequencing data have been deposited in the European Nucle-otide Archive, https://www.ebi.ac.uk/ena (accession nos. ERX3197375–ERX3197410).1C.V.U. and M.G. contributed equally to this work.2Present address: Department of Cell Biology, Embryology, and Genetics, Federal Univer-sity of Santa Catarina, 88040-900 Florianopolis, Brazil.

3Present address: National Council for Scientific and Technological Development, Na-tional Institute of Science and Technology in Molecular Entomology, 21941-902 Rio deJaneiro, Brazil.

4Present address: Division of Microbiology and Parasitology, Department of Pathology,University of Cambridge, CB2 1TN Cambridge, United Kingdom.

5To whom correspondence may be addressed. Email: g.christophides@imperial.ac.uk or d.vlachou@imperial.ac.uk.

This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1919709117/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1919709117 PNAS Latest Articles | 1 of 11

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

June

20,

202

0

midgut subepithelial space encountering the hemolymph thatcarries the complement-like system.The hallmark of the mosquito complement-like system is the

C3-like factor, TEP1 (7, 8). A proteolytically processed form ofTEP1, TEP1cut, circulates in the hemolymph as a complex withLRIM1 and APL1C (9, 10). Upon parasite recognition, TEP1cutis released from the complex and attacks the ookinete, triggeringin situ assembly of a TEP1 convertase that locally processesTEP1 molecules that bind to the ookinete causing lysis and, insome cases, melanization (11). These reactions are regulated byCLIP-domain serine proteases and their inactive homologs (11,12). Ookinete clearance is assisted by actin-mediated cellularresponses of invaded epithelial cells (13).The characterization of Plasmodium falciparum Pfs47 as a

player in parasite evasion of the mosquito complement-like re-sponse has opened new avenues to dissect the mechanisms para-sites employ to endure or indeed evade the mosquito immuneresponse. glycosyl-phosphatidylinositol (GPI)-anchored Pfs47 washypothesized to interfere with activation of JNK signaling and,by doing so, aids ookinetes to escape nitration and subsequentcomplement-mediated attack (14, 15). This parasite immuneevasion function is shared by the Pfs47 ortholog in the rodentmalaria parasite Plasmodium berghei (16), which was earlierthought to be solely involved in fertilization (17).Our transcriptomic profiling of field P. falciparum isolates

from Burkina Faso in the midgut of sympatric Anopheles coluzzii(previously Anopheles gambiae M form) and Anopheles arabiensismosquitoes and a laboratory P. berghei strain in the midgut of A.coluzzii (18) identified hundreds of genes exhibiting conservedand differential expression during gametocyte-to-oocyst develop-ment. Several of them encoding putatively secreted or membrane-associated proteins were made part of a screen to identify genesthat function during parasite infection of the mosquito midgut.These genes were given a candidate gene number preceded by theacronym PIMMS, for “Plasmodium Infection of the MosquitoMidgut Screen.” We previously characterized PIMMS2 that en-codes a subtilisin-like protein involved in midgut traversal (19).Here, we report the characterization of P. falciparum and P. ber-ghei PIMMS43 that encodes a membrane-bound protein found onthe surface of ookinetes and sporozoites. The gene was first reportedin P. berghei to be a target of the transcription factor AP2-O, has arole in mosquito midgut invasion and oocyst development, and wasnamed POS8 (20). A later study by another group reported the geneas being important for ookinete maturation, designating it asPSOP25 (21). Here we demonstrate that PIMMS43 has no de-tectable function in ookinete maturation or mosquito midgutinvasion but plays a key role in ookinete evasion of the mosquitocomplement-like response. We show that disruption of PIMMS43leads to robust complement activation and ookinete eliminationupon completion of midgut traversal and before their trans-formation to oocysts. When the complement system is inactivated,oocyst transformation is restored but sporogony cannot be com-pleted, as the gene is also essential for sporozoite development.Parallel analysis of thousands of African P. falciparum parasitesreveals clear genetic differentiation between populations sampledfrom West or Central and East African countries, inferringparasite adaptation to sympatric vector populations. We furtherdemonstrate that A. coluzzii ingestion of antibodies against P.falciparum PIMMS43 leads to strong inhibition of oocyst devel-opment. The discovery and characterization of PIMMS43 addsto our understanding of parasite immune evasion and malariatransmission through the vector.

Results and DiscussionIdentification of PIMMS43. P. falciparum (PF3D7_0620000) and P.berghei (PBANKA_1119200) PIMMS43 encode deduced proteins of505 and 350 amino acids, respectively. N-terminal signal peptides(amino acids 1 to 25 for PfPIMMS43 and 1 to 22 for PbPIMMS43)

and C-terminal transmembrane domains (amino acids 482 to 504for PfPIMMS43 and 327 to 350 for PbPIMMS43) are predicted forboth proteins. The transmembrane domains are predicted by Pre-dGPI to also contain signals for attachment of a GPI lipid anchorwith 99% probability.PIMMS43 is conserved among species of the Plasmodium

genus. All orthologs are predicted to contain the N-terminalsignal peptide and C-terminal transmembrane domain, as wellas a conserved pair of cysteine residues adjacent to the C ter-minus (SI Appendix, Fig. S1). PbPIMMS43 exhibits a 68% se-quence identity with orthologs in other rodent parasites (i.e.,Plasmodium yoelii and Plasmodium chabaudi) and 27% and 24%with P. falciparum and Plasmodium vivax PIMMS43, respectively.PfPIMMS43 and PvPIMMS43 contain a 60 to 180 nonconservedamino acid insertion with no obvious sequence similarity be-tween them, which are therefore likely to have occurred in-dependently. Another shorter, nonconserved insertion towardthe C terminus of P. vivax and Plasmodium knowlesi PIMMS43includes tandem repeats of Glycine-Serine-Glutamine-Alanine-Serine (GSQAS).

PIMMS43 Transcripts Peak in Ookinetes and Sporozoites. DNAmicroarray profiling of A. coluzzii and A. arabiensis midguts in-fected with P. falciparum field isolates in Burkina Faso revealedthat PfPIMMS43 (referred to in figures as Pfc43) shows pro-gressively increased transcription that peaks 24 h postmosquitoblood feeding (hpbf) (Fig. 1A). These data were corroborated bylaboratory P. falciparum NF54 infections of A. coluzzii using RT-PCR (SI Appendix, Fig. S2A). Low levels of PfPIMMS43 tran-scripts were also detected in in vitro-cultured gametocytes butnot in in vitro-cultured asexual blood stage (ABS) parasites, in-dicating that PfPIMMS43 transcription begins in gametocytesand peaks in zygotes and ookinetes. Transcripts were not de-tected in oocysts 10 d postmosquito blood feeding (dpbf) butreappeared in mosquito salivary glands, indicative of PfPIMMS43reexpression in sporozoites. While we do not detect PfPIMMS43transcripts at 10 dpbf, two published RNA sequencing experimentshave shown that PfPIMMS43 transcripts are detected at 7 (22) and 8dpbf (23) in the midgut oocysts, suggesting that PfPIMMS43 tran-scripts may be too low to be detected at 10 dpbf in our assays.We examined whether the P. berghei PIMMS43 ortholog (re-

ferred to in figures as Pbc43) shows expression profile similar toPfPIMMS43, using quantitative real-time RT-PCR (qRT-PCR)(Fig. 1B) and RT-PCR (SI Appendix, Fig. S2B). In these assays,we used the P. berghei line ANKA507m6cl1 that constitutivelyexpresses GFP (24), hereafter referred to as c507, as well as thenongametocyte producing ANKA 2.33 (NGP) as a control in theRT-PCR assay. The results revealed low levels of PbPIMMS43transcripts in mixed blood stages (MBS) and purified c507 ga-metocytes, which together with absence of transcripts from NGPMBS indicated that PbPIMMS43 transcription begins in gametocytes,similar to PfPIMMS43. Also similar to PfPIMMS43, PbPIMMS43transcript levels were very high 24 hpbf, as well as in purified invitro-produced ookinetes, indicating high PbPIMMS43 tran-scription in ookinetes. Lower transcript levels were detected 2dpbf, presumably due to ookinetes retained in the blood bolusand the midgut epithelium and low-level expression in youngoocysts. No PbPIMMS43 transcripts were detected in matureoocysts 10 dpbf, but strong PbPIMMS43 reexpression was observedin salivary gland sporozoites. Taken together, these data indicatethat P. falciparum and P. berghei PIMMS43 exhibit similar tran-scription patterns starting in gametocytes, peaking in ookinetes,pausing in oocysts, and restarting in salivary gland sporozoites.To investigate PbPIMMS43 protein expression, we raised

rabbit polyclonal antibodies against a codon-optimized fragmentof the protein (amino acids 22 to 327) expressed in Escherichiacoli cells (α-Pbc43opt), and a native protein fragment (aminoacids 22 to 331) expressed in insect Spodoptera frugiperda Sf9

2 of 11 | www.pnas.org/cgi/doi/10.1073/pnas.1919709117 Ukegbu et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

cells (α-Pbc43Sf9). Both recombinant proteins lacked the pre-dicted signal peptide and C-terminal transmembrane domain.We also generated a genetically modified c507 P. berghei line,designated Δc43, where 50% of the PbPIMMS43 coding regionwas replaced with a modified Toxoplasma gondii pyrimethamine-resistance expression cassette (TgDHFR) (SI Appendix, Fig.S3A). Integration of the disruption cassette was confirmed byPCR and pulse field gel electrophoresis (SI Appendix, Fig. S3 Band C). RT-PCR assays confirmed that PbPIMMS43 transcriptscould no longer be detected in gametocytes, ookinetes, and spo-rozoites of the Δc43 line that was henceforth used as a negativecontrol in protein-expression experiments (SI Appendix, Fig. S2B).Western blot analysis was performed in total, triton-soluble

protein extracts prepared under reducing conditions from MBS,purified gametocytes, and in vitro-cultured ookinetes of the c507and Δc43 P. berghei lines (Fig. 1C). Two clear bands of ∼37 and 75kDa were detected in ookinete extracts of the c507 line. Theformer band matches the predicted molecular weight ofPbPIMMS43 monomer and the latter band could correspond toPbPIMMS43 dimer, either a homodimer formed upon disulfidebonding of the conserved pair of cysteine residues or a hetero-dimer. Indeed, under strong reducing conditions, the 75 kDa wasresolved in a single 37-kDa band, whereas under nonreducingconditions only the 75 kDa band could be detected (Fig. 1D). Thisassay was combined with membrane-fractionation of total in vitroookinete extracts, which revealed that both bands were only ob-served in the insoluble fraction and the fraction solubilized bytriton, but not in the soluble (triton nontreated) fraction. Thesedata indicate membrane association of PbPIMMS43, in accor-dance with the prediction of a transmembrane domain anda GPI anchor.We also raised a rabbit polyclonal antibody against a codon-

optimized coding fragment of P. falciparum PIMMS43 (aminoacids 25 to 481) expressed in E. coli cells and lacking the pre-dicted signal peptide and C-terminal transmembrane domain(α-Pfc43opt). We examined the affinity and specificity of thisantibody by generating and using a P. berghei c507 transgenicline (PbPfc43), where PbPIMMS43 was replaced by PfPIMMS43(SI Appendix, Fig. S4A). PCR genotypic analysis confirmedsuccessful modification of the endogenous PbPIMMS43 geno-mic locus (SI Appendix, Fig. S4B), and RT-PCR analysis con-firmed that PfPIMMS43 is transcribed in in vitro-cultured P.berghei ookinetes (SI Appendix, Fig. S4C). Western blot analysisof total protein extracts prepared from purified in vitro-culturedPbPfc43 ookinetes using the α-Pfc43opt antibody revealed a strongband of ∼60 kDa, corresponding to the predicted molecularweight of the deduced PfPIMMS43 protein (SI Appendix, Fig.S4D). This band was absent from c507 and Δc43 protein extracts,confirming the specificity of the α-Pfc43opt antibody. It is note-worthy that, in contrast to what was observed with the PbPIMMS43protein, the results did not show dimerization of the ectopicallyexpressed PfPIMMS43 protein when the analysis was done undernonreducing conditions (SI Appendix, Fig. S4D). While this mayrepresent a true difference between the P. berghei and P. falci-parum PIMMS43 proteins, we cannot rule out the possibility thatPfPIMMS43 dimerization requires, in addition to the conservedcysteine residues, a P. falciparum-specific cofactor that is notpresent in P. berghei and may be related to the nonconserved N-terminal amino acid insertion in PfPIMMS43.

PIMMS43 Protein Is Localized on the Parasite Membrane. We usedthe α-Pfc43opt antibody in indirect immunofluorescence assays toinvestigate the subcellular localization of PfPIMMS43 in P. fal-ciparum NF54 parasite stages. Antibodies against the femalegametocyte and ookinete surface protein Pfs25 and the sporo-zoite surface protein PfCSP (Circumsporozoite protein) were usedas stage-specific controls. The results showed that PfPIMMS43prominently localizes on the surface of female gametes or early-stage zygotes found in the A. coluzzii blood bolus 1 hpbf, as well ason the surface of ookinetes traversing the mosquito midgut epi-thelia and sporozoites found in the mosquito salivary gland lumenat 25 hpbf and 16 dpbf, respectively (Fig. 2A). There was no evi-dence of α-Pfc43opt antibody staining of in vitro-cultured asexualblood stage or gametocytes, suggesting that expression of PfPIMMS43protein starts after fertilization. No signal was detected with theα-Pfc43opt rabbit preimmune serum that was used as a negativecontrol.Immunofluorescence assays of P. berghei c507 and control

Δc43 parasite stages using the α-Pbc43opt antibody revealed that,similarly to its P. falciparum ortholog, PbPIMMS43 localizes onthe surface of A. coluzzii midgut-traversing ookinetes and salivary

Fig. 1. PIMMS43 transcription profiles and protein expression. (A) DNAmicroarray transcriptional profiling of Pfc43 in A. coluzzii (Ac) and A. ara-biensis (Aa) midguts. Bars show transcript abundance at indicated timepoints relative to 1 hpbf and are average of three biological replicates. Errorbars show SEM. (B) Relative abundance of Pbc43 transcripts in blood stages,in vitro ookinetes, and A. coluzzii mosquito stages, as determined by qRT-PCR in the c507 line and normalized against the constitutive expressed GFP.Each bar is the average of three biological replicates. Error bars show SEM.(C) Western blot analysis under reducing conditions (3% [vol/vol] 2-mer-capthoethanol) using the α-Pbc43opt antibody on whole-cell lysates of c507parasites. Pbc43 protein bands are indicated with asterisks. Δc43 parasiteswere used as a negative control. GFP was used as a loading control. (D)Western blot analysis under reducing (Left; 5% [vol/vol] 2-mercapthoethanol)and nonreducing (Right) conditions using the α-Pbc43opt antibody on frac-tionated in vitro ookinetes. Pbc43 protein bands are indicated with asterisks.Δc43 ookinetes were used as a negative control. P28 and GFP were used asstage-specific and loading controls, respectively. Soluble (Sol), Triton soluble(Tri Sol), and insoluble (Insol) fractions are shown. Abbreviations: Gc(−), non-activated gametocytes; Gc(+), activated gametocytes; Ook, ookinetes.

Ukegbu et al. PNAS Latest Articles | 3 of 11

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

June

20,

202

0

gland sporozoites (Fig. 2B). For the control Δc43 line, which asreported below does not develop beyond the ookinete stage,sporozoites were obtained from infections of LRIM1 knockdown(KD) mosquitoes (see below). Like PfPIMMS43, despite thepresence of transcripts, no signal was detected in gametocytes.Furthermore, no PbPIMMS43 signal was detected in early-stagezygotes present in the blood bolus 1 hpbf, suggesting that trans-lation starts later during ookinete development. In both species,the protein was detectable on the surface of 2-d-old oocysts foundon the A. coluzzii midgut cell wall and reappeared in sporozoitesfound in mature P. falciparum oocysts 11 dpbf and P. bergheioocysts 15 dpbf (SI Appendix, Fig. S5).

PIMMS43 Deletion Mutants Fail to Reach Oocyst-Stage Infection. Wephenotypically characterized the P. berghei Δc43 line generatedas described above. Consistent with the PbPIMMS43 expressiondata, Δc43 parasites exhibited normal development in mouse bloodstages (SI Appendix, Fig. S6). Both male gametocyte activation, asmeasured by counting exflagellation centers (Fig. 3A), andmacrogametocyte-to-ookinete conversion rate, both in vitro and inthe A. coluzziimidgut lumen (Fig. 3B), were comparable to the c507parental line, indicating that no developmental defects are accom-panying the parasite gametocyte-to-ookinete developmental transi-tion. However, no oocysts were detected in A. coluzzii midguts at 3,5, 7, or 10 dpbf, indicating complete abolishment of oocyst forma-tion (Fig. 3C and SI Appendix, Table S1). Thus, oocyst and salivarygland sporozoites were never observed, and transmission to micefollowing mosquito bite-back was abolished (SI Appendix, Table S2).

To validate the specificity of this phenotype, we reintroducedPbPIMMS43 into the Δc43 locus by replacing the TgDHFR genecassette with the PbPIMMS43 coding sequence flanked by its 5′and 3′ untranslated regions (UTRs) and followed by the humanDHFR gene cassette (SI Appendix, Fig. S7A). Successful in-tegration was confirmed with PCR (SI Appendix, Fig. S7B).Phenotypic characterization of the resulting Δc43::c43wt parasiteline in A. coluzzii infections showed that oocyst development wasfully restored (SI Appendix, Fig. S7C and Table S1).These data were in disagreement with those reported previ-

ously, which showed that PSOP25 knockout (KO) parasites exhibitreduced ookinete conversion rates and defective ookinete mat-uration (21). To investigate this discrepancy, we generated anew PIMMS43 KO (Δc43red) line in the 1804cl1 (c1804) P. ber-ghei line that constitutively expresses mCHERRY (25), using thesame disruption vector (PbGEM-042760) as the one used by theauthors of the previous study, which leads to 74% removal of thegene coding region (SI Appendix, Fig. S8 A and B). Phenotypicanalysis showed that Δc43red parasites show normal ookineteconversion rates both in vitro and in A. coluzzii infections butproduced no oocysts (SI Appendix, Fig. S8C), a phenotypeidentical to that of the Δc43 line. Similar results were obtained ininfections of Anopheles stephensi, the vector of choice in theprevious studies (SI Appendix, Fig. S8D). Interestingly, the numberof oocysts in A. stephensi infections was very small but not zero.This is consistent with the findings by Kaneko et al. (20), as well aswith the general understanding that the A. stephensi Nijmegenstrain, which was genetically selected for high susceptibility toparasite infections (26), has a less robust immune response than

Fig. 2. PIMMS43 protein localization. (A) Immunofluorescence assays of P. falciparum NF54 parasites found in mosquito blood bolus of at 1 hpbf (Left),ookinetes traversing the mosquito midgut epithelium at 25 hpbf (Center), and salivary gland sporozoites at 16 dpbf (Right), stained with α-Pfc43opt (green)and the female gamete/zygote/ookinete α-Pfs25 (red) or sporozoite α-PfCSP (purple) antibodies. DNA was stained with DAPI. Staining with preimmune serumwas used as a negative control. (B) Immunofluorescence assays of P. berghei 507 early sexual stages (activated gametocytes and/or early zygotes) in mosquitoblood bolus at 1 hpbf (Left), ookinetes traversing the mosquito midgut epithelium at 25 hpbf (Center), and salivary gland sporozoites at 21 dpbf (Right),stained with α-Pbc43opt (green), female gamete/zygote/ookinete surface α-P28 (red) or sporozoite surface α-PbCSP (purple) antibodies. DNA was stained withDAPI. Staining of the Δc43 parasite with α-Pbc43opt was used as a negative control. Note that Δc43 sporozoites were obtained from infections of LRIM1 kdmosquitoes. Images are de-convoluted projection of confocal stacks. BF, bright field. (Scale bars, 5 μm.)

4 of 11 | www.pnas.org/cgi/doi/10.1073/pnas.1919709117 Ukegbu et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

A. coluzzii. Nonetheless, no sporozoites were detected in the A.stephensi midgut 15 dpbf (SI Appendix, Fig. S8E).

Ookinetes Lacking PIMMS43 Are Killed by the Mosquito Complement-like Response upon Midgut Traversal. We examined whether thePIMMS43 KO phenotype was due to defective ookinete motilityand, hence, capacity to invade or traverse the mosquito midgutepithelium. Ookinete motility assays showed that Δc43 ookinetesmoved on Matrigel with average speed that was not significantlydifferent from c507 ookinetes (Fig. 3D).Next, a potential defect in midgut epithelium invasion and

traversal was assessed in infections of A. coluzzii where CTL4 (C-type lectin 4) was silenced by RNA interference. CTL4 KD leadsto melanization of ookinetes at the midgut subepithelial spaceupon epithelium traversal providing a powerful means to visualizeand enumerate ookinetes that successfully traverse the midgut

epithelium. The number of Δc43 melanized ookinetes was compa-rable to that of the c507 line that was used as control (Fig. 3E and SIAppendix, Table S3), indicating that Δc43 ookinetes successfullytraverse the midgut epithelium but fail to transform to oocysts.A similar phenotype was previously reported for P47 KO

parasites that are eliminated by mosquito complement-like re-sponses upon emergence at the midgut subepithelial space (16).To examine whether the same applies to Δc43 parasites, we in-fected A. coluzzii mosquitoes in which genes encoding two majorcomponents of the complement-like system, TEP1 and LRIM1,were individually silenced. Enumeration of oocysts 10 dpbf, andcomparison with control mosquitoes injected with LacZ double-stranded RNA, revealed that Δc43 oocyst development waslargely restored in both TEP1 and LRIM1 KD mosquitoes (Fig.3F); although the numbers of recovered Δc43 oocysts were still

Fig. 3. Phenotypic analysis of P. berghei Δc43 KO mutant parasites. Male gametocyte activation measured as percentage of exflagellating male gametocytes(A) and female gamete conversion to ookinetes in vitro (Left), and in vivo in the A. coluzziimidgut of (Right) (B) of c507 wt and Δc43 parasites. Error bars showSEM. (C) Δc43 oocyst development at 3, 5, 7, and 10 dpbf in A. coluzzii. ***P < 0.0001, Mann–Whitney U test. (D) Speed of c507 wt and Δc43 ookinetesmeasured from time-lapse microscopy, captured at one frame/5 s for 10 min. Red lines indicate mean and error bars show SEM. ns, not significant. (E) Melanizedookinete numbers in CTL4 kd A. coluzzii infected with c507 wt and Δc43 parasite lines. Red lines indicate median; ns, not significant; n, number of biologicalreplicates. (F) Effect of LRIM1 and TEP1 silencing on Δc43 oocyst numbers in A. coluzzii midguts. dsLacZ-injected mosquitoes were used as controls. Red linesindicate median; n, number of independent experiments; ***P < 0.0001, Mann–Whitney U test. (G) Representative images of rescued Δc43 oocysts in LRIM1 KDmosquitoes showing variable morphology and smaller size compared to c507 wt oocysts. (Scale bars, 30 μm.) (H) Box plot of diameter measurements of Δc43 andc507 wt oocysts at 14 and 16 dpbf. Upper and lower whiskers represent the largest and smallest oocyst diameter, respectively. Horizontal line in each box indicatesmean of 2 biological replicates and whiskers show SEM. N is number of oocysts; *P < 0.05, and ***P < 0.0001 using unpaired Student’s t test.

Ukegbu et al. PNAS Latest Articles | 5 of 11

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

June

20,

202

0

lower than the numbers of WT (c507) oocysts in TEP1 andLRIM1 KD mosquitoes (SI Appendix, Table S4).Taken together, these data indicate that the absence of

PIMMS43 does not affect the capacity of ookinetes to invade andtraverse the mosquito midgut epithelium but instead required forprotection from or evasion of the mosquito immune response. Theobservation that Δc43 oocyst numbers are still inferior to WTparasite oocyst numbers in both TEP1 and LRIM1 KDmosquitoescould suggest that immune responses additional to TEP1-mediatedkilling of Δc43 ookinetes. Indeed, it has been previously shown thatsome dead ookinetes in the midgut epithelium are not bound byTEP1 (7), indicating alternative means employed by the mos-quito to kill Plasmodium ookinetes. Other mosquito immunefactors, such as fibrinogen-related proteins (FREPs or FBNs)and LRRD7, are also important for midgut infection (27, 28). Ofthese, FBN9 is shown to colocalize with ookinetes in the midgutepithelium, probably mediating their death (28). Any suchmechanism employed by the mosquito to kill Δc43 ookineteswould have to be TEP1-independent. Since TEP1 attack is po-tentiated by prior marking of ookinetes by effector reactions ofthe JNK pathway (5, 6), it is plausible that Δpbc43 ookinetes areexcessively marked for death either by the same mechanismobserved for Pfs47-null mutants or an independent mechanism.Alternatively, due to the spatial and temporal limitations of RNAi,it is possible that residual activity of TEP1 and LRIMI retained inthe KD mosquitoes is responsible for the incomplete rescue ofΔc43 oocyst numbers. Nonetheless, all of the above scenariossuggest that PIMMS43, like P47, directly or indirectly interfereswith the mosquito immune response promoting ookinete survival.

PIMMS43 Is Additionally Required for Oocyst Maturation andSporozoite Development. We observed that the rescued Δc43oocysts in LRIM1 or TEP1 KD mosquitoes were morphologi-cally variable and smaller in size compared to c507 oocysts (Fig.3G). At 14 and 16 dpbf the average Δc43 oocyst diameter was20.1 and 17.2 μm, compared to 27.4 and 30.9 μm of c507 oo-cysts, respectively (Fig. 3H). All pairwise comparisons werestatistically significant and revealed that the mean Δc43 oocystdiameter at 16 dpbf was smaller than 14 dpbf, indicating pro-gressive degeneration of Δc43 oocysts. In addition, Δc43 oocystsin LRIM1 KD mosquitoes yielded a very small number ofmidgut and salivary gland sporozoites compared to c507 oo-cysts, and the ratio of salivary gland to midgut sporozoites wassignificantly smaller for Δc43 compared to control c507 parasites(SI Appendix, Table S6). Furthermore, the few Δc43 sporozoitesthat reached the salivary glands of in LRIM1 KD mosquitoescould not be transmitted to mice through mosquito bites.These data suggested that Δc43 parasites are defective not only

with respect to ookinete toleration of the mosquito complement-like response but also with sporozoite development and infectivity.We investigated whether bypassing midgut invasion, a process inwhich ookinetes are marked for elimination by complement-likereactions, could rescue Δc43 sporozoite development and trans-mission to a new host. In vitro-produced Δc43 and control c507ookinetes were injected into the hemocoel of A. coluzzii mosqui-toes, and sporozoites found in the mosquito salivary glands 21d later were enumerated. The results revealed that no Δc43 spo-rozoites could be detected in the mosquito salivary glands, andconsequently, mosquitoes inoculated with Δc43 ookinetes couldnot transmit malaria to mice, in contrast to mosquitoes inoculatedwith c507 ookinetes (SI Appendix, Table S6). While these dataargue for an additional, essential function of PbPIMMS43 insporozoite development, complement-independent effects directlyimpacting the injected ookinete may be responsible for the spo-rozoite phenotype observed in this experiment.Next, we investigated whether PfPIMMS43 could complement

the function of its P. berghei ortholog by infecting naïve A.coluzzii mosquitoes with the PbPfc43 parasite line and counting

the number of oocysts detected in the mosquito midguts. Infec-tions with c507 and Δc43 parasites served as positive and negativecontrols, respectively. The results showed that the PbPfc43 lineexhibited an intermediate phenotype compared to c507 and Δc43both in terms of both infection prevalence and intensity (SI Ap-pendix, Fig. S4E and Table S4). Oocysts were morphologicallyvariable and smaller in size compared to c507 oocysts and con-tained a small number of sporozoites that could not be transmittedto a new mouse host (SI Appendix, Table S5), resembling thephenotype obtained with Δc43 infections following silencing of themosquito complement-like system. We examined whether thispartial complementation phenotype could be affected uponLRIM1 silencing. Indeed, a significant increase in both the in-fection prevalence and oocyst numbers was observed (SI Appendix,Fig. S4E and Table S4), yet oocysts remained small and mor-phologically variable and produced few sporozoites (SI Appendix,Table S5). These results suggest that PfPIMMS43 can only partlycomplement the function of its PbPIMMS43 ortholog and cor-roborate the dual function of PIMMS43 in ookinete to oocysttransition and in oocyst maturation and sporozoite development,respectively.

PIMMS43 KO Leads to Compromised Ookinete Fitness and Attack bythe Complement-Like Response. We carried out RNA next-generation sequencing of P. berghei Δc43 and c507 infected A.coluzzii midguts at 1 and 24 hpbf to further investigate the Δc43phenotype during mosquito midgut infection. P. berghei and A.coluzzii transcriptomes were processed separately, and compar-atively analyzed at each time point for each parasite line(Datasets S1 and S2). Three independent biological replicatesand three technical replicates for each biological replicate wereperformed.At 1 hpbf, when asexual parasite stages and gametocytes are

sampled from the mosquito blood bolus, almost all 17 changesregistered between Δc43 and c507 parasites concerned genesbelonging to multigene families (pir, fam-a, and fam-b) and ri-bosomal RNAs (SI Appendix, Fig. S9A). We hypothesize that thisis due to differential expression of such genes between clonalparasite lines rather than differences related to the disruption ofPbPIMMS43. As many as 163 genes were differentially regulatedbetween the Δc43 and c507 parasites at 24 hpbf, of which 137were down-regulated (41 at least twofold) and 26 were up-regulated (9 at least twofold) (Fig. 4A). Gene ontology analysisrevealed several biological processes and three cellular compo-nent terms that were significantly enriched in the differentiallyregulated gene set (SI Appendix, Table S7). All gene ontologyterms were related to host–parasite interactions, includingmicronemal secretion, entry into host cell, and parasite move-ment. Genes included in this list encode known ookinete se-creted or membrane-associated proteins—such as CTRP, SOAP,MAEBL, WARP, PLP3-5, PIMMS2, HADO, PSOP1, PSOP7,PSOP26, GAMA (also known as PSOP9)—and others, all ofwhich were down-regulated in Δc43 parasites. The expression ofthe oocyst capsule protein Cap380 gene that begins in ookineteswas also affected (29).These data could be explained by a smaller ratio of ookinetes

to other parasite stages sampled from the midgut at 24 hpbf inΔc43 infections compared to c507 infections. Although the datafrom the ookinete melanization assays showed that differencesbetween Δc43 and c507 in ookinete numbers exiting the mos-quito midgut were not statistically significant (P = 0.0947), thesedifferences were almost twofold both with regards to median andarithmetic mean (SI Appendix, Table S3). This difference couldjustify the observed twofold down-regulation of genes showingenriched expression in ookinetes. A second hypothesis is that Δc43parasites exhibit deficient expression of genes involved in ookinetesecretions and movement. The latter hypothesis is less appealing,as it is difficult to explain how absence of a membrane-associated

6 of 11 | www.pnas.org/cgi/doi/10.1073/pnas.1919709117 Ukegbu et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

protein without obvious signaling domains could affect the tran-scription of all other genes. However, the two hypotheses are notmutually exclusive, and both indicate that disruption of PIMMS43leads to compromised ookinete fitness.Analysis of A. coluzzii midgut transcriptional responses to in-

fection by Δc43 compared to c507 identified 192 and 122 differ-entially regulated genes at 1 and 24 hpbf, respectively (DatasetS2). At 1 hpbf, 154 (88 over twofold) genes were down-regulatedand 38 (21 over twofold) were up-regulated (SI Appendix, Fig.S9B). However, these genes did not appear to follow any func-tional pattern, and annotation enrichment analyses did not yieldany significant results. In contrast, at 24 hpbf, and although thenumber of identified genes was smaller (109 down-regulated, 71over twofold; 13 up-regulated, 5 over twofold), most genes shownto date to be involved in systemic immune responses of thecomplement-like system and downstream effector reactions—in-cluding TEP1, LRIM1, APL1C, and various clip-domain serineprotease homologs—were down-regulated (Fig. 4B). Enrichmentanalysis confirmed that the serine protease/protease/hydrolase andthe serine protease inhibitor/protease inhibitor protein classeswere significantly overrepresented in this gene list. When considered

together with the increased complement activity observedagainst Δc43 compared to the c507 ookinetes, these data couldsuggest induction of a negative feedback mechanism to down-regulate this self-damaging innate immune response. However,most of these genes are thought to be largely, and in some casesexclusively, expressed in hemocytes and fat body cells; therefore,their detection as down-regulated in midgut tissues cannot be easilyexplained. Thus, a more possible explanation is that midgut in-fection by Δc43 ookinetes causes mobilization and differentiation ofhemocytes attached to the midgut tissues, as shown previously (30–32), causing a temporal depletion of relevant transcripts from themidgut tissue.We examined this hypothesis by measuring the abundance of

transcripts encoding the three major components of the complement-like system, TEP1, LRIM1, and APL1C, in the midgut and wholebody (excluding legs, wings, and heads) of A. coluzzii mosquitoesinfected with Δc43 or control c507 parasites at 24 hpbf. Since theΔc43 phenotype was similar to the Δpbp47 phenotype (16), andbecause preliminary data indicated similar A. coluzzii midgutresponses to the two mutant parasite lines, transcript abundancein infections with Δpbp47 parasites were also examined. Theresults revealed a striking difference in transcript abundance ofall three genes between midgut and whole mosquitoes (SI Ap-pendix, Fig. S10). In accordance with the RNA-sequencing data,the relative transcript abundance in infections with the twomutant parasite lines compared to control infections was lower inthe midgut but higher in whole mosquitoes. These data areconsistent with our hypothesis that ookinetes lacking PIMMS43or P47 trigger hemocyte mobilization and consequent depletionin the midgut tissue, although the up-regulation of the threegenes in whole mosquitoes could also be explained by up-regulation in other immune tissues, such as the fat body.

PIMMS43 Exhibits Significant Geographic Structure Among African P.falciparum Populations. It has been shown that Pfs47 presentsstrong geographic structure in natural P. falciparum populations,both between continents and across Africa (33–35). Further-more, a small-scale genotypic analysis of oocysts sampled fromA. gambiae and Anopheles funestus mosquitoes in Tanzaniarevealed significant differentiation in Pfs47 haplotypes sampledfrom the two vectors (36). These data are consistent with naturalselection of Pfs47 haplotypes by the mosquito immune systemand a key role of this interaction in parasite–mosquito co-evolution (34). However, a different study showed that poly-morphisms in the Pfs47 locus alone could not fully explain theobserved variation in complement-mediated immune evasion ofAfrican P. falciparum strains (37).We investigated the genetic structure of African P. falciparum

populations with regards to PfPIMMS43, and compared this tothe structure of Pfs47, using a rich dataset of 1,509 genome se-quences of parasites sampled from 11 African countries in thecontext of the P. falciparum Community Project (http://www.malariagen.net/). The PfPIMMS43 analysis revealed significantpopulation differentiation as determined by the Fixation Index(FST), mostly between populations of some West or Central(Democratic Republic of the Congo, DC) and East Africancountries (FST > 0.1) (Fig. 5). The highest FST is detected incomparisons of Ugandan, DC, or Kenyan populations with WestAfrican populations. The most differentiated single nucleotidepolymorphisms (SNPs) are detected within the nonconservedregion that is unique to P. falciparum (Dataset S3). Within thisregion, an SNP that leads to the nonsynonymous substitution ofSerine-217 to Leucine (S217L) is highly differentiated betweensampled Kenyan/Tanzanian and all other populations, while anearby SNP that leads to substitution of Glutamate-226 to Lysine(E225K) has swept to almost fixation in Ugandan populations.The PIMMS43 FST profile does not fully match the FST profile

of Pfs47 that also presents strong genetic differentiation between

Fig. 4. P. berghei and A. coluzzii gene expression 24 hpbf. (A) Volcano plotof P. berghei gene expression in Δc43 vs. c507 wt parasite lines in the A.coluzzii midgut. (B) Volcano plots of A. coluzzii midgut transcriptional re-sponses to Δc43 vs. c507 wt parasites. The x axes show log2 fold-change and yaxes show log10 P value calculated using one-way ANOVA. Blue and orangefilled circles indicate genes that are at least twofold down down-regulatedand twofold up-regulated, respectively. Black circles show with no signifi-cant differential regulation. Known gene names are indicated.

Ukegbu et al. PNAS Latest Articles | 7 of 11

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

June

20,

202

0

West and East Africa but is particularly strong for populationssampled in Madagascar and Malawi versus West African and DCpopulations (SI Appendix, Fig. S11). The most highly differenti-ated SNPs are within domain 2 (D2) of the protein (Dataset S3).An SNP leading to substitution of Leucine-240 to Isoleucine(L240I) is almost fixed in Madagascar and Ugandan versus WestAfrican populations, while a nearby SNP leading to the non-synonymous substitution of Asparagine-271 to Isoleucine (N271I)is highly prevalent in DC versus all other populations, especiallythose sampled from East Africa. Our analysis also detected allfour SNPs previously shown to differentiate between African(NF54) and New World (GB8) P. falciparum laboratory lines andlead to amino acid substitutions in the D2 region that contributeto immune evasion (38); however, these SNPs were neither highlyprevalent nor did they present significant geographic structureapart from that leading to Isoleucine-248 substitution to Leucineor Valine (I258L/V) that is significantly prevalent (FST > 0.1) insampled Ugandan populations. These data concur with the hy-pothesis presented previously that polymorphisms in the D2 regionof Pfs47, even those leading to synonymous substitutions, can alterthe parasite immune evasion properties (38). Finally, one of thesubstitutions defining the East versus West African differentiation isthat of Glutamate-27 to Aspartate (E27D) at the start of themature protein. This SNP is almost fixed in sampled Madagascarpopulations.Together, these data reveal that PfPIMMS43 and Pfs47 exhibit

significant geographic structure, consistent with their deducedrole in parasite immune evasion. They also suggest that differentselection pressures are exerted on each of these genes, whichconcurs with the hypothesis that the two proteins serve different

functions. A major difference between West and East Africanvector species is the presence of both A. gambiae (A. gambiae S-form) and A. coluzzii (A. gambiae M-form) in West Africa butonly A. gambiae in East Africa. Interestingly, a resistant allele ofTEP1, TEP1rB, is shown to have swept to almost fixation in WestAfrican A. coluzzii but be absent from A. coluzzii sampled fromCameroon, consistent with the high PfPIMMS43 FST observedbetween Central and West African parasite populations, as wellas from all sampled A. gambiae populations (39). Therefore, it istempting to speculate that a difference between West and EastAfrican vectors in their capacity to clear parasite infectionsthrough complement responses may have contributed to theobserved PfPIMMS43 and Pfs47 genetic structure.Moreover, A. funestus and A. arabiensis appear to have re-

cently taken over from A. gambiae as the primary malaria vectors inmany areas of East Africa (40), in contrast to West Africa where A.gambiae and A. coluzzii remain the primary vectors. While nothing isknown about the capacity of A. funestus to mount complement-likeresponses against malaria parasites, A. arabiensis is shown to be aless good vector of P. berghei but can be transformed into ahighly susceptible vector, equal to A. gambiae, when its com-plement system is silenced (41). Finally, Anopheles merus is onlyfound in coastal East Africa; although its abundance and con-tribution to malaria transmission has been increasing (42) it isunlikely that it has majorly contributed to structuring parasitepopulations.

Incubation with Anti-PIMMS43 Antibodies Significantly ReducesMalaria Transmission by A. coluzzii. We examined in both P. falci-parum and P. berghei whether targeting PIMMS43 using anti-bodies generated against each of the respective orthologousproteins could reduce parasite infectivity and malaria transmissionpotential. For P. falciparum transmission-blocking assays, purifiedIgG α-Pfc43opt antibodies were added to gametocytemic blood atfinal concentrations of 0, 50, 125, and 250 μg/mL prior to offeringthis as bloodmeal to female A. coluzzii mosquitoes through opti-mized standard membrane feeding assays (SMFAs) (43). Oocystspresent in mosquito midguts at day 7 postfeeding were enumer-ated. The results showed strong inhibition of both infection intensityand infection prevalence in an antibody dose-dependent manner(Fig. 6A and SI Appendix, Table S8). At 125 and 250 μg/mL ofantibody following four biological replicates, the overall inhibitionof infection intensity observed was 57.1% and 76.2%, and theoverall inhibition of infection prevalence was 37.3% and 35.6%,respectively (P < 0.0001).Similar results were obtained with P. berghei transmission upon

addition of α-Pbc43Sf9 antibodies to blood drawn from infectedmice and provided to mosquitoes as bloodmeal in SMFAs. Sta-tistically significant inhibition of both infection intensity andprevalence was detected at all antibody concentrations tested (i.e.,50, 100, and 250 μg/mL) in an antibody dose-dependent manner(Fig. 6B and SI Appendix, Table S9). At 100 μg/mL, the inhibitionof oocyst intensity was 72.7% and the inhibition of infectionprevalence was 35.5%, and these values increased to 90.3% and65.6% at 250 μg/mL, respectively (P < 0.0001).A recent study has shown that antibodies binding a 52-amino

acid region of Pfs47 confer strong transmission blocking of lab-oratory P. falciparum strains in A. gambiae (44). In the samestudy, antibodies binding different regions of the protein showedeither weak or no transmission-blocking activity, consistent withan earlier study reporting that none of three monoclonal anti-bodies against Pfs47 could affect P. falciparum infections in A.stephensi (45). These findings agree with the general un-derstanding that antibodies binding different regions of a tar-geted protein can have profound differences in their blockingactivity, especially when antibodies have a primarily neutralizingfunction (46, 47). Indeed, our polyclonal α-Pbc43opt antibodyraised against codon-optimized PbPIMMS43 expressed in E. coli

Fig. 5. Population genetic analysis of PfPIMMS43 in African P. falciparum.PfPIMMS43 FST values of 1,509 P. falciparum populations sampled from pa-tients across Africa (Upper) and schematic representation of SNPs with highFST values leading to amino acid substitutions in each deduced protein(Lower). (Upper) Color coding indicates comparisons between countries inWest, Central, and East Africa. Central Africa includes populations sampledonly from the DC. White bars overlaid with colored bars indicate the FST ofPfs47. (Lower) Boldfaced amino acid substitutions are those deriving fromSNPs with total FST > 0.1, and the rest of the substitutions are those showinghigh FST in comparison between populations sampled from specific coun-tries. Substitutions marked with red stars are those showing very high FSTand have swept to almost fixation in some populations. Yellow spikes showthe positions of conserved cysteine residues. Burkina Faso, BF; DemocraticRepublic of the Congo, DC; Gambia, GM; Ghana, GH; Guinea, GN; Kenya, KE;Madagascar, MG; Malawi, MW; Mali, ML; Tanzania, TZ; Uganda, UG.

8 of 11 | www.pnas.org/cgi/doi/10.1073/pnas.1919709117 Ukegbu et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

cells appeared to not exhibit any transmission-blocking activityagainst P. berghei, despite producing strong signals in Western blotanalyses and immunofluorescence assays. However, antibodiesagainst fragments of PSOP25 (synonym of PIMMS43) expressed inE. coli cells have been previously shown to inhibit P. berghei in-fection in A. stephensi (21, 48), albeit not as strongly as ourα-Pbc43Sf9 antibodies.

Concluding Remarks and Perspectives. We demonstrate that PIMMS43is required for parasite evasion of the mosquito immune response, arole also shared by P47 in both P. falciparum and P. berghei (14,16). The mechanism by which these molecules exert theirfunction is unclear. A general explanation may lie with theirGPI constituents or with their structural role in the formationof the ookinete sheath. On the one hand, Plasmodium GPIs areknown to modulate the vertebrate host immune system (49),and studies have shown that mosquitoes mount a specific im-mune response against GPIs (50, 51). On the other hand, theintegrity of the ookinete sheath may be important for coun-teracting attacks by or acting as molecular sinks of free radicalsproduced during traversal of midgut epithelial cells (5, 6) or bydirectly enduring the attack of the complement-like system.Ookinetes lacking such membrane proteins may thus be irreversiblydamaged and subsequently eliminated by the mosquito complement-like response. In relation to this, a specific function could beattributed to the conserved cysteine residues present in theseproteins. Apart from their role in forming disulphide bridges,thus serving a structural purpose, the ability of cysteine thiolgroups to regulate the redox potential may be relevant (52).Interestingly, midgut infection with P. berghei is shown to inhibitthe expression of catalase that mediates the removal of free rad-icals, and silencing catalase exacerbates ookinete elimination (53).Nonetheless, population genetic analyses indicate a more specificrole of the two proteins in parasite–mosquito interactions andcoadaptation.Notwithstanding their exact function in parasite immune

evasion, PIMMS43, P47, and possibly other proteins involved inparasite immune evasion are targets of interventions aiming toblock malaria transmission in the mosquito. One such approachis transmission-blocking vaccines aiming at generating antibodiesin the human serum which, when ingested by mosquitoes to-gether with gametocytes, interfere with the function of theseproteins and block transmission to a new host (54). Several pu-tative transmission-blocking vaccines are currently being in-vestigated at a preclinical stage, including those targeting thegametocyte and ookinete proteins Pfs230, Pfs48/45, and Pfs25(55). Another, more ambitious approach is the generation ofgenetically modified mosquitoes expressing single-chain anti-bodies or nanobodies that bind these proteins conferring re-fractoriness to infection and leading to malaria transmissionblocking (56, 57). Such genetic features can be spread within wildmosquito populations in a super-Mendelian fashion via means ofgene drive (e.g., CRISPR/Cas9) and can lead to sustainable localmalaria elimination (58).

Fig. 6. P. falciparum and P. berghei transmission blocking with anti-PIMMS43 antibodies. Transmission-blocking efficacies of anti-PIMMS43 an-tibodies on (A) P. falciparum and (B) P. berghei infections of A. coluzziishown as dot plots of oocyst number distribution (Upper) and forest plots ofgeneralized linear mixed-model analysis (Lower). The α-Pfc43opt andα-Pbc43Sf9 antibodies were provided through SMFAs at concentrations of 50,125, and 250 μg/mL, and 50, 100, and 250 μg/mL, respectively, and comparedwith no antibodies and UPC10 antibodies that were used as negative con-trols for P. falciparum and P. berghei, respectively. Individual data pointsrepresent oocyst numbers from individual mosquitoes at 7 and 10 dpbf from

two/four and three biological SMFA replicates with P. falciparum and P.berghei, respectively. m/M are mean/median oocyst infection intensities, alsoshown as horizontal blue and red lines, respectively. IP, oocyst infectionprevalence; N, number of midguts analyzed; n, number of independentexperiments; ns, not significant. Statistical analysis was performed withMann–Whitney U test for infection intensity and Fisher’s exact test for in-fection prevalence; **P < 0.005; ***P < 0.0001. In generalized linear mixed-model analyses, the variation of fixed-effect estimates for each replicate(squares) and all replicates (diamonds) are shown (±95% confidence interval,glmmADMB). The square size is proportional to the sum of midguts analyzedin each replicate. *P < 0.05; ***P < 0.0001.

Ukegbu et al. PNAS Latest Articles | 9 of 11

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

June

20,

202

0

Materials and MethodsEthics Statement. All animal procedures were reviewed and approved by theImperial College Animal Welfare and Ethical Review Body and carried out inaccordance with the Animal Scientifics Procedures Act 1986 under the UnitedKingdomHome Office licenses PLL70/7185 and PPL70/8788. Human red bloodcells provided by the National Blood Service of the United Kingdom NationalHealth Service were obtained from healthy donors upon taking a writteninformed consent.

Protein Expression and Purification. Recombinant PfPIMMS43 and PbPIMMS43were expressed in E. coli (PfPIMMS43 and PbPIMMS43) and S. frugiperda insectcells (PbPIMMS43). Fragments without the signal peptide and the C-terminalhydrophobic domain were expressed as Histidine-fusion proteins. Details ofthe cloning, expression, and purification are found in SI Appendix, Supple-mentary Materials and Methods.

Generation of Transgenic Parasites. Two transgenic parasites carrying 50%and 74% KO of the P. berghei PIMMS43 coding DNA sequence were createdby homologous recombination in the P. berghei c507 and 1804cl1 lines re-spectively. For the generation of P. berghei transgenic parasites expressingP. falciparum PIMMS43, the PfPIMMS43 coding sequence was expressedunder the control of the PbPIMMS43 5′UTR and 3′UTR and at the originalPbPIMMS43 locus in the P. berghei c507 line. For the complementation as-says, we reintroduced the full length PbPIMMS43 in the Δc43 line under thecontrol of its 5′UTR and 3′UTR and at its original locus. Details of the crea-tion of these transgenic parasites are found in SI Appendix, SupplementaryMaterials and Methods.

RNA-Sequencing. Total RNA was isolated from midguts infected with WT andPbPIMMS43 KO parasites at 1 and 24 hpbf. The RNA was used to preparecDNA libraries (New England Biolabs Ultra prep kit) and run on an Illumina

HiSEq. 3000 instrument. For details, see SI Appendix, Supplementary Mate-rials and Methods.

Transmission-Blocking Assays. Polyclonal antibodies generated against theabove recombinant proteins were utilized in transmission-blocking assays.Here, mosquitoes were fed with P. berghei and P. falciparum by SMFAs in thepresence of the generated polyclonal antibodies. Oocyst infection intensityand oocyst infection prevalence was determined at 10 dpbf. See SI Appen-dix, Supplementary Materials and Methods.

Other Materials andMethods.Materials andmethods for RT-PCR and qRT-PCR,Western blot analysis, immunofluorescence assays, phenotypic assays, andpopulation genetic analysis are described in detail in in SI Appendix, Sup-plementary Materials and Methods.

Data Availability Statement. The RNA-sequencing data are available throughthe European Nucleotide Archive with experiment codes ERX3197375 toERX3197410.

ACKNOWLEDGMENTS. The authors thank Ana-Rita Gomes for assistancewith transfection and cloning of the Δc43red parasite line; Melina Campos forassistance with generalized linear mixed-model analysis; and Katarzyna Sala,Chrysanthi Taxiarchi, Lara Selles, and Neil Mac Aogain for technical assis-tance. The work was funded by Wellcome Trust Investigator Award107983/Z/15/Z (to G.K.C.); Wellcome Trust Project Grant 093587/Z/10/Z (toG.K.C. and D.V.); and Bill and Melinda Gates Foundation GrantOPP1158151 (to G.K.C.). L.D.P.R. was funded by a Royal Society Newton In-ternational Fellowship NF161472. A.M.B. was funded by Medical ResearchCouncil New Investigator Grant MR/N00227X/1. Microscopy was performedin the Facility for Imaging by Light Microscopy (FILM) at Imperial CollegeLondon.

1. A. M. Clayton, Y. Dong, G. Dimopoulos, The Anopheles innate immune system in the

defense against malaria infection. J. Innate Immun. 6, 169–181 (2014).2. M. Povelones, M. A. Osta, G. K. Christophides, The complement system of malaria

vector mosquitoes. Adv. Insect Physiol. 51, 223–242 (2016).3. Y. Alavi et al., The dynamics of interactions between Plasmodium and the mosquito: A

study of the infectivity of Plasmodium berghei and Plasmodium gallinaceum, and

their transmission by Anopheles stephensi, Anopheles gambiae and Aedes aegypti.

Int. J. Parasitol. 33, 933–943 (2003).4. R. C. Smith, J. Vega-Rodríguez, M. Jacobs-Lorena, The Plasmodium bottleneck: Ma-

laria parasite losses in the mosquito vector. Mem. Inst. Oswaldo Cruz 109, 644–661

(2014).5. L. S. Garver, G. de Almeida Oliveira, C. Barillas-Mury, The JNK pathway is a key me-

diator of Anopheles gambiae antiplasmodial immunity. PLoS Pathog. 9, e1003622

(2013).6. Gde. A. Oliveira, J. Lieberman, C. Barillas-Mury, Epithelial nitration by a peroxidase/

NOX5 system mediates mosquito antiplasmodial immunity. Science 335, 856–859

(2012).7. S. Blandin et al., Complement-like protein TEP1 is a determinant of vectorial capacity

in the malaria vector Anopheles gambiae. Cell 116, 661–670 (2004).8. E. A. Levashina et al., Conserved role of a complement-like protein in phagocytosis

revealed by dsRNA knockout in cultured cells of the mosquito, Anopheles gambiae.

Cell 104, 709–718 (2001).9. M. Fraiture et al., Two mosquito LRR proteins function as complement control factors

in the TEP1-mediated killing of Plasmodium. Cell Host Microbe 5, 273–284 (2009).10. M. Povelones, R. M. Waterhouse, F. C. Kafatos, G. K. Christophides, Leucine-rich re-

peat protein complex activates mosquito complement in defense against Plasmodium

parasites. Science 324, 258–261 (2009).11. M. Povelones et al., The CLIP-domain serine protease homolog SPCLIP1 regulates

complement recruitment to microbial surfaces in the malaria mosquito Anopheles

gambiae. PLoS Pathog. 9, e1003623 (2013).12. H. Yassine, L. Kamareddine, S. Chamat, G. K. Christophides, M. A. Osta, A serine

protease homolog negatively regulates TEP1 consumption in systemic infections of

the malaria vector Anopheles gambiae. J. Innate Immun. 6, 806–818 (2014).13. T. Schlegelmilch, D. Vlachou, Cell biological analysis of mosquito midgut invasion: The

defensive role of the actin-based ookinete hood. Pathog. Glob. Health 107, 480–492

(2013).14. A. Molina-Cruz et al., The human malaria parasite Pfs47 gene mediates evasion of the

mosquito immune system. Science 340, 984–987 (2013).15. U. N. Ramphul, L. S. Garver, A. Molina-Cruz, G. E. Canepa, C. Barillas-Mury, Plasmo-

dium falciparum evades mosquito immunity by disrupting JNK-mediated apoptosis of

invaded midgut cells. Proc. Natl. Acad. Sci. U.S.A. 112, 1273–1280 (2015).16. C. V. Ukegbu et al., Plasmodium berghei P47 is essential for ookinete protection from

the Anopheles gambiae complement-like response. Sci. Rep. 7, 6026 (2017).17. M. R. van Dijk et al., Three members of the 6-cys protein family of Plasmodium play a

role in gamete fertility. PLoS Pathog. 6, e1000853 (2010).

18. K. A. Akinosoglou et al., Characterization of Plasmodium developmental tran-scriptomes in Anopheles gambiae midgut reveals novel regulators of malaria trans-mission. Cell Microbiol. 17, 254–268 (2015).

19. C. V. Ukegbu, K. A. Akinosoglou, G. K. Christophides, D. Vlachou, Plasmodium bergheiPIMMS2 promotes ookinete invasion of the Anopheles gambiae mosquito midgut.Infect. Immun. 85, e00139 (2017).

20. I. Kaneko, S. Iwanaga, T. Kato, I. Kobayashi, M. Yuda, Genome-wide identification ofthe target genes of AP2-O, a Plasmodium AP2-family transcription factor. PLoSPathog. 11, e1004905 (2015).

21. W. Zheng et al., Functional characterization of Plasmodium berghei PSOP25 duringookinete development and as a malaria transmission-blocking vaccine candidate.Parasit. Vectors 10, 8 (2017).

22. E. Gómez-Díaz et al., Epigenetic regulation of Plasmodium falciparum clonally variantgene expression during development in Anopheles gambiae. Sci. Rep. 7, 40655 (2017).

23. G. Zanghì et al., A specific PfEMP1 is expressed in P. falciparum sporozoites and playsa role in hepatocyte infection. Cell Rep. 22, 2951–2963 (2018).

24. C. J. Janse et al., High efficiency transfection of Plasmodium berghei facilitates novelselection procedures. Mol. Biochem. Parasitol. 145, 60–70 (2006).

25. T. Annoura et al., Two Plasmodium 6-Cys family-related proteins have distinct andcritical roles in liver-stage development. FASEB J. 28, 2158–2170 (2014).

26. A. M. Feldmann, T. Ponnudurai, Selection of Anopheles stephensi for refractorinessand susceptibility to Plasmodium falciparum. Med. Vet. Entomol. 3, 41–52 (1989).

27. Y. Dong et al., Anopheles gambiae immune responses to human and rodent Plas-modium parasite species. PLoS Pathog. 2, e52 (2006).

28. Y. Dong, G. Dimopoulos, Anopheles fibrinogen-related proteins provide expandedpattern recognition capacity against bacteria and malaria parasites. J. Biol. Chem.284, 9835–9844 (2009).

29. P. Srinivasan, H. Fujioka, M. Jacobs-Lorena, PbCap380, a novel oocyst capsule protein,is essential for malaria parasite survival in the mosquito. Cell Microbiol. 10, 1304–1312(2008).

30. J. C. Castillo, A. B. B. Ferreira, N. Trisnadi, C. Barillas-Mury, Activation of mosquitocomplement antiplasmodial response requires cellular immunity. Sci. Immunol. 2,eaal1505 (2017).

31. C. Frolet, M. Thoma, S. Blandin, J. A. Hoffmann, E. A. Levashina, Boosting NF-kappaB-dependent basal immunity of Anopheles gambiae aborts development of Plasmo-dium berghei. Immunity 25, 677–685 (2006).

32. R. C. Smith, C. Barillas-Mury, M. Jacobs-Lorena, Hemocyte differentiation mediatesthe mosquito late-phase immune response against Plasmodium in Anopheles gam-biae. Proc. Natl. Acad. Sci. U.S.A. 112, E3412–E3420 (2015).

33. M. Manske et al., Analysis of Plasmodium falciparum diversity in natural infections bydeep sequencing. Nature 487, 375–379 (2012).

34. A. Molina-Cruz, G. E. Canepa, C. Barillas-Mury, Plasmodium P47: A key gene formalaria transmission by mosquito vectors. Curr. Opin. Microbiol. 40, 168–174 (2017).

35. A. Molina-Cruz et al., Plasmodium evasion of mosquito immunity and global malariatransmission: The lock-and-key theory. Proc. Natl. Acad. Sci. U.S.A. 112, 15178–15183(2015).

10 of 11 | www.pnas.org/cgi/doi/10.1073/pnas.1919709117 Ukegbu et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

36. T. G. Anthony, S. D. Polley, A. P. Vogler, D. J. Conway, Evidence of non-neutralpolymorphism in Plasmodium falciparum gamete surface protein genes Pfs47 andPfs48/45. Mol. Biochem. Parasitol. 156, 117–123 (2007).

37. M. Eldering et al., Variation in susceptibility of African Plasmodium falciparum ma-laria parasites to TEP1 mediated killing in Anopheles gambiae mosquitoes. Sci. Rep. 6,20440 (2016).

38. G. E. Canepa, A. Molina-Cruz, C. Barillas-Mury, Molecular analysis of pfs47-mediatedPlasmodium evasion of mosquito immunity. PLoS One 11, e0168279 (2016).

39. B. J. White et al., Adaptive divergence between incipient species of Anophelesgambiae increases resistance to Plasmodium. Proc. Natl. Acad. Sci. U.S.A. 108, 244–249(2011).

40. E. W. Kaindoa et al., Interventions that effectively target Anopheles funestus mos-quitoes could significantly improve control of persistent malaria transmission insouth-eastern Tanzania. PLoS One 12, e0177807 (2017).

41. T. Habtewold, Z. Groom, G. K. Christophides, Immune resistance and tolerance strategiesin malaria vector and non-vector mosquitoes. Parasit. Vectors 10, 186 (2017).

42. N. Cuamba, C. Mendis, The role of Anopheles merus in malaria transmission in an areaof southern Mozambique. J. Vector Borne Dis. 46, 157–159 (2009).

43. T. Habtewold et al., Streamlined SMFA and mosquito dark-feeding regime signifi-cantly improve malaria transmission-blocking assay robustness and sensitivity. Malar.J. 18, 24 (2019).

44. G. E. Canepa et al., Antibody targeting of a specific region of Pfs47 blocks Plasmo-dium falciparum malaria transmission. NPJ Vaccines 3, 26 (2018).

45. B. C. van Schaijk et al., Pfs47, paralog of the male fertility factor Pfs48/45, is a femalespecific surface protein in Plasmodium falciparum. Mol. Biochem. Parasitol. 149, 216–222 (2006).

46. J. S. Armistead et al., Antibodies to a single, conserved epitope in Anopheles APN1inhibit universal transmission of Plasmodium falciparum and Plasmodium vivax ma-laria. Infect. Immun. 82, 818–829 (2014).

47. P. J. Bustamante et al., Differential ability of specific regions of Plasmodium falcipa-rum sexual-stage antigen, Pfs230, to induce malaria transmission-blocking immunity.Parasite Immunol. 22, 373–380 (2000).

48. W. Zheng et al., Identification of three ookinete-specific genes and evaluation of theirtransmission-blocking potentials in Plasmodium berghei. Vaccine 34, 2570–2578 (2016).

49. R. S. Naik et al., Glycosylphosphatidylinositol anchors of Plasmodium falciparum:Molecular characterization and naturally elicited antibody response that may provideimmunity to malaria pathogenesis. J. Exp. Med. 192, 1563–1576 (2000).

50. R. B. Arrighi, I. Faye, Plasmodium falciparum GPI toxin: A common foe for man andmosquito. Acta Trop. 114, 162–165 (2010).

51. J. Lim, D. C. Gowda, G. Krishnegowda, S. Luckhart, Induction of nitric oxide synthasein Anopheles stephensi by Plasmodium falciparum: Mechanism of signaling and therole of parasite glycosylphosphatidylinositols. Infect. Immun. 73, 2778–2789 (2005).

52. L. B. Poole, The basics of thiols and cysteines in redox biology and chemistry. FreeRadic. Biol. Med. 80, 148–157 (2015).

53. A. Molina-Cruz et al., Reactive oxygen species modulate Anopheles gambiae immu-nity against bacteria and Plasmodium. J. Biol. Chem. 283, 3217–3223 (2008).

54. R. E. Sinden, “Infection of mosquitoes with rodent malaria” in The Molecular Biologyof Insect Disease Vectors, J. M. Crampton, C. B. Beard, C. Louis, Eds. (Springer, 1997),pp. 67–91.

55. N. Chaturvedi, P. K. Bharti, A. Tiwari, N. Singh, Strategies & recent development oftransmission-blocking vaccines against Plasmodium falciparum. Indian J. Med. Res.143, 696–711 (2016).

56. V. M. Gantz et al., Highly efficient Cas9-mediated gene drive for population modi-fication of the malaria vector mosquito Anopheles stephensi. Proc. Natl. Acad. Sci.U.S.A. 112, E6736–E6743 (2015).

57. A. T. Isaacs et al., Engineered resistance to Plasmodium falciparum development intransgenic Anopheles stephensi. PLoS Pathog. 7, e1002017 (2011).

58. R. Carballar-Lejarazú, A. A. James, Population modification of Anopheline species tocontrol malaria transmission. Pathog. Glob. Health 111, 424–435 (2017).

Ukegbu et al. PNAS Latest Articles | 11 of 11

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

June

20,

202

0